Continual Vision-Language-Action Learning on LIBERO single-task learning scenario
0.98FARCRL-VLA (Q)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CRL-VLA (Q)Value Formulation=Q-based goal-conditioned value2026.02 | 0.98 | -0.02 | -0.06 | 3 | |
| MTL2026.02 | 0.96 | 0.11 | -0.06 | 6 | |
| LWF2026.02 | 0.67 | -0.5 | -0.06 | 50 | |
| CRL-VLA (V)Value Formulation=V-based goal-conditioned value2026.02 | 0.67 | -0.49 | -0.06 | 50 | |
| ER2026.02 | 0.6 | -0.51 | -0.06 | 53 | |
| SL2026.02 | 0 | -0.62 | -0.5 | 62 |